The tumor suppressor p53 promotes carcinoma invasion and collective cellular migration

Summary Loss of function of the tumor suppressor p53 is generally thought to increase cell motility and invasiveness. Using 2-D confluent and 3-D spheroidal cell motility assays with bladder carcinoma cells and colorectal carcinoma cells, we report, to the contrary, that loss of p53 can decrease cell motility and invasion. Abstract For migration of the single cell studied in isolation, loss of function of the tumor suppressor p53 is thought to increase cell motility. Here by contrast we used the 2-D confluent cell layer and the 3-D multicellular spheroid to investigate how p53 impacts dissemination and invasion of cellular collectives. We used two human carcinoma cell lines, the bladder carcinoma EJ and the colorectal carcinoma HCT116. We began by replicating single cell invasion in the traditional Boyden chamber assay, and found that the number of invading cells increased with loss of p53, as expected. In the confluent 2-D cell layer, however, for both EJ and HCT, speeds and effective diffusion coefficients for the p53 null types compared to their p53 expressing counterparts were significantly smaller. Compared to p53 expressers, p53 null cells exhibited more organized cortical actin rings together with reduced front-rear cell polarity. Furthermore, loss of p53 caused cells to exert smaller traction forces upon their substrates, and reduced formation of cryptic lamellipodia. In a 3-D collagen matrix, p53 consistently promoted invasion of the multicellular spheroids into surrounding matrix. Together, these results show that p53 expression in these carcinoma model systems increases collective cellular migration and invasion. As such, these studies point to paradoxical contributions of p53 in single cell versus collective cellular migration.


Introduction
Among human cancers, the tumor suppressor p53 is the most mutated gene and serves not only as an 44 inducer of cancer cell senescence and apoptosis [1,2], but also as a central suppressor of cancer cell 45 migration and metastasis [3][4][5][6]. For example, in 3-dimensional (3D) Matrigel assays, loss of p53 increases 46 single cell invasion by enhancing cell contractility [7][8][9][10]. In wound healing assays, p53 can decrease the 47 migration distance of leading cells by the inhibition of epithelial-mesenchymal transition (EMT) [11]. In Boyden chamber assay [7-10]. It is now recognized, however, that metastatic disease is dominated by 52 collective cellular migration rather than single cell migration [14][15][16]. In the case of collective cellular 53 migration, the cell-cell interactions can be quite strong and highly cooperative [17][18][19][20][21]. Moreover, the 54 cellular collective can become jammed, immobile, and solid-like, or unjammed, mobile, and fluid-like 55 [18,22,23]. In the case of single cell migration, by contrast, none of these potent mechanisms are operative. 56 It remains unclear, however, how p53 functions in the context of such collective phenomena. 57 To address that issue, here we studied migration and invasion in 2D confluent cell layers and 3D 58 multicellular spheroids. Two human cell lines were used, the bladder carcinoma EJ and the colorectal 59 carcinoma HCT116. We first replicated single cell invasion assays in the Boyden chamber and found 60 results consistent with previous studies [7 -9]; loss of p53 increased the invasion of the single carcinoma 61 cell. To our surprise, however, loss of p53 in either EJ or HCT 116 cells suppressed cellular dissemination 62 in 2-D confluent cell layers. In that case, loss of p53 was associated with reduced lamellipodia formation 4 63 and weaker cell-substrate interactions. To better mimic tumor biology we also conducted studies using 3D 64 multicellular spheroids embedded in collagen matrix. We found these results in the 3D multicellular 65 spheroidal assay to be consistent with the 2-D confluent assay. These results, taken together, demonstrate 66 paradoxical contributions of p53 in single cell versus collective cell migration.

68
In 2D confluent cell layers, p53 increases collective cellular motility 69 To determine the function of p53 in collective cell motility, we used both gain and loss of p53 70 function in colorectal and bladder carcinoma cell lines: stable wild type (p53 +/+ ) and stable p53 null (p53 -/-) 71 HCT 116 and Tet-off inducible EJ cell line (Methods). In EJ cell line, p53 knocking out (EJ p53 off) was 72 established by the addition of doxycycline to the culture media. We began by replicating assays of single 73 cell invasion in the Matrigel-coated Boyden chamber as reported in previous studies [7-10], and found 74 consistent results; loss of p53 increased cell invasion (125±53 versus 415±101 cells per well for EJ, p = 75 0.009 249±65 versus 891±239 for HCT 116, p = 0.03) (Fig. S1). 76 We then went on to assays of cell migration in the 2-D confluent cell layer. The confluent cell layer 77 was cultured on 1.2 kPa polyacrylamide gel, and red fluorescent beads embedded in the gel surface. We          tumor suppressor p53 promotes the carcinoma spheroid to invade a larger area (Fig. 6). These results 202 demonstrate that the tumor suppressor p53 promotes carcinoma cell escape from their neighbors and more 203 efficient invasion into matrix. 205 Our results show that p53 promotes the dissemination and invasion of the cellular collectives.

206
Compared to p53 expressing counterparts, the p53 null carcinomas shared the features of having highly 207 organized rings of cortical F-actin, and more rounded and less polarized cell shape (Fig. 3B and Fig. 4B).

208
These results are consistent with the general consensus that loss of p53 promotes cellular rounding [7,30].  To our knowledge these studies are the first to quantify the mechanical effects of p53 on cell-substrate 219 interaction. Measurements in both EJ and HCT 116 suggest that the loss of p53 consistently causes these 220 carcinoma cells to exert smaller traction forces on their substrate (Fig. 2). It remains unclear, however, if 221 reduction in traction forces might be attributable to reduced lamellipodia formation and less motility. Many studies suggest that p53 can prevent epithelial-mesenchymal transition (EMT) and increase E-225 cadherin expression to decrease cancer cell motility [11,26,34,35]. Nevertheless, at least one study [9]

285
3D multicellular spheroid assay 286 A 200l cell suspension (2×10 5 cells/ml for EJ p53 on, 5×10 5 cells/ml for HCT 116) was cultured in an 287 ultra-low attachment 96-well plate (VWR, 29443-034) to form the multicellular spheroid. After 48h, each 288 spheroid was carefully pipetted into 1.5mg/ml collagen matrix (Advanced Biomatrix, 5005). Cell culture 289 medium was used to adjust the collagen concentration, and the collagen matrix was equilibrated through 290 10X PBS (volume ratio, 1:10 between 10X PBS and collagen) and 1M NaOH (0.5% of total matrix 291 volume). These processes were performed on ice to avoid collagen polymerization. We then moved the 292 collagen matrix into 37°C incubator to induce collagen polymerization. After 1h, we monitored the 293 invasion of the spheroids via Leica microscope (Leica DMI8).

294
Immunofluorescence and confocal laser-scanning microscopy 295 For immunofluorescence analysis of cell lines, cell layers from the 2-D assays were fixed in 4% 296 paraformaldehyde/PBS for 10 min, permeabilized and blocked in 0.2% Triton X-100/PBS containing milk 297 for 20 min. The cell layers were stained with primary E-cadherin antibody (Invitrogen, 334000) [11], then  previous studies [11,38]. The 24-well Boyden chamber (Corning, 354480) contains an 8m pore size PET membrane which has 316 been coated by Matrigel. We added warmed serum-free culture medium into the interiors of the chambers 317 and the bottoms of the wells, and allowed the system to rehydrate for 2 hours in humidified tissue culture 318 incubator. After rehydration, we carefully removed the medium without disturbing the layer of Matrigel. 319 We prepared the serum-free cell suspension (5×10 4 cells/ml for EJ, 2×10 5 cells/ml for HCT 116), and then 320 added 0.5ml into the chamber. We used sterile forceps to transfer the chambers to the wells containing 321 0.75ml culture medium containing serum as chemoattractant. Cells were incubated in the chambers for 22 322 hours in 37°C, 5% CO 2 incubator. After the incubation, we used cotton tipped swabs to scrub the surface 323 of the chamber twice to remove the non-invading cells from the upper surface. Cells were then fixed and 324 stained F-actin and nucleus via the above methods. We counted the nuclei of the entire well bottom 325 through the particle analysis in Image-J. N=4 for each type from two experiments.